low power design and wifi

No, It's ZigAnt. Yes, pay me $1 per year.

We have it on a 16K Atmega169, 64K PIC32MX130F064 client and a 512K PIC32MX675F512 server, which has 10M/100M ethernet.

WiFi vs 802.15.4

Not an Apples to Apples comparison.

Are there any web pages that compare only WiFi models ?

Yes, Bad (for OP's app) RF vs. Good RF.

OP did not require 801.11 only.

802.11 needs at least 500mW to operate and around 100 seconds to connect. Doesn't matter what models you pick.

I am not a WiFi expert, but is that 100 seconds typical ?

I am sure, but I'll ask, does distance require more power or will the router not accept the connection if power is too low ? ( or does it slow down )

nd)

f

t. Doesn't matter what models you pick.

Yes, but it depends on many other factors as well. Since Dot 11 (802.11) r equires full handshake before transmission (think TCP), You have to keep th e radio active for the entire period. OTOH, Dot 15 can send independent pa cket (think UDP) in half duplex. ZagAnt (#1) use 10 seconds turn around fu ll duplex, for applications that needs bi-directional communications. For comparisons:

Mode System Power RF Power Minimum Time Dot 11 F 500mW 100mW 100s Dot 11 H N/A N/A N/A Dot 15 F 80mW 10mW All (Server/Gateway mode) Dot 15 F 80mW 1mW 10s (Client Bi-directional) Dot 15 H 80mW 1mW 1s (Client Uni-directional)

Note #1: Someone already claimed ZigAnt. So, we are switching to ZagAnt. Ant for A symetric Network Transmission protocol.

It will either drop the data rate or more retries (at 500ms intervals).

I have read many papers and more than happy to show your more numbers.

100 seconds? No way.

A few hundred milliseconds - yes, this I believe. A second or two in many cases, yes.

From a cold standing start, a WiFi client has to find the access point (locate the correct frequency, and "hear" a beacon with the correct SSID) and then go through several back-and-forth transmissions in order to "associate" with the AP and (in many cases) negotiate the security protocol. Since most WiFi access points only transmit a beacon every 100 milliseconds, this puts a lower limit on how fast a new association can be set up. If you're using a heavy-weight security protocol which requires certificate validation, or a trip to a Radius server, it'll slow things down even further.

If the client and AP have already successfully associated with one another, and the client has gone into a low-power "sleep" mode, the client can usually awaken and start transmitting again within one beacon cycle (100 MS) I believe.

Rather than WiFi i think i would try 902 to 928 MHz spread spectrum (DSSS) ISM radio. Of course for both cases bigger buffers help rather than hurt, there is sync-up time even for DSSS or FHSS. GMSK might be a better idea.

?-)

What is the transmitter duty cycle, i.e. how many bytes do you need to send every 30 s ?

The higher the peak transmission rate, the higher received power requirement (look at the receiver sensitivity specs for various data rates) and hence more transmission power is needed (typically limited to 100 mW or +20 dBm by standards).

This also requires a stiff power supply in order to deliver the peak power without too much voltage sag. A huge capacitor might be needed in parallel with a small battery. If the mobile device also needs a receiver, the power consumption can be considerable, especially if the receiver needs to be kept on all the time. If it is possible to keep the Rx power on for only during some short mutually agreed periods, the average consumption can be reduced.

What about sending the data at a lower rate constantly. Assuming some ISM device with 9600 bit/s data rate, it can transfer about 1 kilobyte/s or 30 kB in 30 s. Is that enough ?

This has the advantage of reducing the peak current and hence a battery without a storage capacitor should be enough.

It should be noted that in a free space environment, the path loss (actually loss of receiver antenna capture area) is increased by 6 dB, each time the RF frequency is doubled. Thus, using omnidirectional antennas, the received signal power at 433 MHz ISM band is 15 dB stronger compared to 2.45 GHz with the same transmitter power.

This advantage could be used to lower the Tx power on increase the range 5-6 times in free space. The greater diffraction at lower frequencies will help to propagate the signal around the obstacles.

For that 50 m range required, a few mW of Tx power should be enough.

Thus, you should make a careful comparison between burst mode WiFi and continuous ISM band operation.

This is our setup, similar to what the OP needs:

The biggest cap we have is a 47uF 0805 SMD chip.

Data: 10 bytes max. Refresh rate: 5 seconds min. Transmission packet: 64 bytes max. Range: 20 meters

Carrier: 2.4GHz Data rate: 256Kb/s

Transmission window: 64 * 8 / 256,000 = 2 msec Transmitter ram-up delay = 5 msec

Transmitter system power = 200mW Transmitter RF power = 1mW

Transmitter battery = 800mAHr Battery cycle = 72 Hours

// receiver is main connected, power does not matter much Receiver system power = 300mW Receiver RF power = 10mW